During medical imaging, patient motion is virtually unavoidable presenting a significant source of image degradation and artifacts, which in turn, can impact the quality of care received by many patients. One method used to address patient motion is to employ visual tracking systems (VTS), which involves placing a small number of reflective markers on the patient that can be located and tracked in 3D space using stereo cameras. The tracked markers can then be used as a surrogate to a structure of interest, such as the heart. The tracked surrogate provides a quantifiable measure that can be directly correlated with the motion of the target structure, which can then be used for prospective or retrospective motion correction. However, the markers add complexity to patient workflow, provide only a sparse sampling of the patient's surface, and the optimal number and placement of markers is unknown. This project addresses the shortcomings of marker-based VTS systems by proposing to develop and test a novel low-cost marker-less VTS system for tracking patient motion within a hybrid imaging system. The primary barrier for successfully tracking the patient's body surface using marker-less VTS is that clothing prevents an unobstructed view of the patient's body. The candidate, Dr. Lindsay, is proposing to incorporate a translational method using computational approaches to compensate for garments within the proposed marker-less VTS system. Dr. Lindsay has received training in computer science and computer vision and previously has focused on the computational aspects of capturing, modeling, and rendering of physical phenomena such as light and is proposing to extend this knowledge to the medical imaging field. Dr. Lindsay's goals for this proposal are to acquire the necessary training and skills that will allow him to translate his expertise and existing computational skills to an area f research with medical relevance: namely, detecting and tracking patient motion for the purpose of improving motion correction for hybrid- imaging. Dr. Lindsay's long-term career goal is to become an accomplished independent investigator focusing on solving significant problems in medical imaging by translating advances in Computer Science. The proposed work will take place the University of Massachusetts Medical Center (UMass) in Worcester, MA under the primary mentorship of Dr. Michael King, an internationally known expert in the area of medical imaging and medical physics in Nuclear Medicine. Drs. Gennert, Sullivan, Licho have expertise in computer vision, mechanical engineering and medical imaging, and radiology and Nuclear Medicine, respectively will also serve as co-mentors.
The specific aims of the proposed project are to: 1) develop a novel marker-less VTS, which we call PT- Cam, using low-cost using state-of-the-art camera technology, 2) model surface motion as a surrogate for internal motion of the heart, and 3) conduct tests of the PT-Cam system, in order to determine the acceptability and feasibility of clinical usage, on patients undergoing clinical MPI PET/CT imaging. Thus, the main objective of this program of research is to develop a clinically viable method for motion tracking patients undergoing hybrid-imaging studies by using a novel marker-less surface-tracking method that can compensate for clothing and provide modeling of interior motion of structures from surface tracking. If successful, not only will this project provide an innovative, marker-less VTS system for low-cost surface tracking for motion compensation during hybrid imaging but it will also give the candidate the necessary training needed to become an independent investigator in the medical field and potentially a leader in the field of medical imaging.
It is estimated that 7.2 million people per year die world-wide from coronary artery disease (CAD), and myocardial perfusion imaging (MPI) has become a critical tool for screening, diagnosis, prognosis, and monitoring treatment of CAD. During medical imaging such as MPI, patient motion is virtually unavoidable and presents a significant source of image degradation and it has been suggested that upwards of 40% of cardiac studies in PET are effected by some type of motion with similar findings for SPECT. The proposed project will directly address the shortcomings of previous methods for tracking patient motion by developing and testing a novel low-cost marker-less visual tracking system for patient motion within hybrid imaging.
